Google| Subject | Re: Lenses and sharpening |
| From | Eric Stevens |
| Date | 09/18/2014 05:16 (09/18/2014 15:16) |
| Message-ID | <c7ck1a1vpl08l37ceikbofgr5q7pl12ar3@4ax.com> |
| Client | |
| Newsgroups | rec.photo.digital |
| Follows | Sandman |
| Followups | Sandman (4h & 53m) |
On 17 Sep 2014 09:46:23 GMT, Sandman <mr@sandman.net>wrote:
SandmanAll of which is true, but it's not what Floyd meant.
In article <ka1i1a56i1u0u9fr6gljrjqv3cgfkm7lau@4ax.com>, Eric Stevens wrote:SandmanSavageduckEric Stevens
It seems that you have never worked with a truly non-destructive workflow, with Photoshop and Lightroom I have a totally reversible workflow which can deal with reverting crops, spot removal, content aware fill, content aware move, any of the various grad filters available, and filters, including the notorious USM.
The reason that all this argument is underway is that you and nospam fail to recognise that a "totally reversible work flow" is one thing but a reversible process is another.
Not really, no.
reversible adjective able to be reversed, in particular: • (of the effects of a process or condition) capable of being reversed so that the previous state or situation is restored
That fits both scenarios. Floyd is stubbornly trying to force everyone else to have "reversible" mean that a process needs another process to reverse its effect, but that's not how the word works. Floyd is notoriously ignorant about word meanings, so no surprise there.
The undo function in Photoshop makes any process reversible, simple as that. If you want to be able to save the file and reverse USM when opening it again, use smart filters. That means that USM is 100% reversible at any point in the future, on any image.Eric StevensSandman
What Floyd has been saying is that sharpening with a high-pass filter is basically the same as Gaussian blur except that one goes forward and the other goes backwards. Whatever you do with one can be undone with the other. This is not the same as just cancelling the operation as you do when you delete it from a sidecar file.
Of course it's not the same. That doesn't mean that deleting an instruction that leads to a specific result doesn't mean the instruction is reversible. The fact that you *can* delete it means that it is by definition reversible.
In fact, "Reversible" comes from the word "reverse", which means "move back". The undo function is the most obvious example of a reversible process.
Deleting (or turning off) an instruction in a non-destructive workflow reverses its affect on the result.
I had hoped point to Wikipedia but their article on 'reversible process' starts off by assuming a knowledge of entropy and then goes on from there. To avoid this, I will have to be technical.
The idea of a 'reversible process' emerged from the study of thermodynamics in the mid 19th century.
Everyone (?) knows that air is compressible and can be reduced in volume by putting it in a cylinder and squeezing it into a smaller volume. Most people have had the experience of doing this with a bicycle pump and discovering it becomes too hot to hold.
When you compress air you are doing work on it - putting energy into the air - and this manifests itself by heat (that's why the bicycle pump gets hot). When you heat air it tries to expand and if you restrain it, it's pressure goes up instead. If you quickly compress air (at an initial pressure P0) with a piston in a cylinder, the pressure will rise (to a pressure, say P2) and it will get hot. If you wait a while the cylinder and the air will cool down (to a pressure P1) and the pressure in the cylinder will fall to a lower level.
Now, if you compress the air from P0 very slowly, the heat of compression will leak out as fast as it is generated and the pressure will eventually rise to P1 without having got any higher. All the heat of compression will have leaked away into the surroundings and the compressed air will be at their temperature. This is called isothermal compression.
Alternatively, if you compress the air very quickly, or if the cylinder is very well insulated, the heat of compression cannot escape and the air will be compressed to P2. This is called adiabatic compression.
The pressure P2 at the end of adiabatic compression will be higher than the pressure P1 reached at the end of isothermal compression.
Consider the isothermal cylinder in which the air is compressed and is still at ambient temperature. Having compressed the air, you can now let it expand by moving the piston back down. The expansion of the air will cause the temperature to drop and the air will shrink, with the result that when the piston reaches the bottom the air will not be at it's initial state but at less than the initial pressure P0. However if the piston is lowered very slowly the cooling of the expanding air will be negated by heat leaking in from the surroundings. Hence when the expansion has ceased the air will still be at ambient temperature and will be back at the initial pressure P0.
Now consider the adiabatic cylinder which is at a higher temperature and pressure than was the isothermal cylinder. No heat leaked out through the cylinder walls when the air was compressed and none is required to flow back in when the air is caused to expand by the piston moving down. In this case the air will be back at it's initial state without requiring any heat flow to or from the outside.
The adiabatic expansion is a fully reversible process because it does not require any input from the outside for it to be brought to any of it's possible states..
The isentropic process is not fully reversible as it cannot be brought to any of it's possible states without input (heating or cooling) from the outside.
Consideration of all of this lead via Willard Gibbs to the concept of entropy, which fortunately I don't have to explain other than to say it is measure of the energy lost in a thermodynamic process. The energy lost in an adiabatic process is zero and for this reason they are often called isentropic.
The idea of entropy has been applied to information theory. In this case it is not energy which is lost but information. This extends to the consideration of the information contained in digital images.
If an image can be transformed from one state to another without loss of information, the transformation is isentropic and the transformation process is fully reversible.
If the transformation of an image from one state to another entails the loss of information then it cannot be brought back to it's initial state without the provision of the information which has been lost. Such a transformation process is not fully reversible.
What Floyd was saying was that High Pass Filter sharpening and Gaussian Blur are basically the same process and that process is fully reversible. He was also saying that Unsharp Mask is not fully reversible and does involve the loss of information.
None of this has got anything to do with the use of the sidecar files used by Lightroom, DxO, NX2, NX-D, Darktable or any other application. --
Regards,
Eric Stevens
09/13/2014-
